Skip to content

Naive dumb simple TensorFlow like framework (academic toy project)

License

Notifications You must be signed in to change notification settings

letalvoj/matrix-flow

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MatrixFlow

This is a dumb simple naive implementation of TensorFlow like framework for matrices. I implemented it as a tiny toy project to learn how a back-propagation work. I tried to keep the code as Scala-ish as possible.

Workshop TODOs

  • caching forward pass
  • parallelization (SDG)
  • probabilistic programming (inspiration)

Possible improvements

Those are just two must have ideas (for me to not to forget):

  • the forward pass should be run only once and cached for the backward pass to be efficient.
  • create interface for Op such that an explicit function for derivatives of all parameters can be specified
  • might be useful when implementing more sophisticated numerical methods...
  • ...

Licence MIT

About

Naive dumb simple TensorFlow like framework (academic toy project)

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages